Researcher Giovanna Luisa Hirata dos Anjos, a member of SMARTNESS, received the Best Paper Award at the Undergraduate and Scientific Initiation Workshop held during the Brazilian Symposium on Computer Networks and Distributed Systems (SBRC) 2026.
The award-winning paper, entitled “Intrusion Detection in DNS Traffic Using Neural Networks and Explainability,” proposes an intrusion detection system for DNS traffic capable of identifying threats such as malware, phishing, and spam using neural networks.
According to the researcher, the study’s main scientific contribution was the auditing of the model through explainability techniques using SHAP. This approach made it possible to detect and eliminate a hidden bias in the dataset. After correcting this structural flaw, the model was able to identify attacks based on their actual behavioral and lexical characteristics.
The results demonstrated robust performance, achieving an average F1-score of 0.95 for legitimate traffic and 0.82 for malicious traffic. To ensure reproducibility, all project code was made publicly available as open-source software.
The recognition received at SBRC 2026 highlights the relevance of the research and its potential contribution to the development of artificial intelligence-based cybersecurity solutions.